AuNUs have great potential in SERS because of their sharp tips and unique optical properties. However, the number of tips, their sizes, and their locations relative to the cores of the AuNUs are hard to control.26In addition, it is difficult to achieve the effective adsorption of probe molecules on the rough and tipped surfaces. Because of these factors, it has been doubtful whether they would be good as SERS substrates. In this regard, we attempted to control and optimize several factors that affect the Raman enhancement in AuNUs. In this study, we did not consider temperature and pH and set them as constant in all the conditions because the temperature can affect nanoparticle shape and R6G is not pH sensitive.27We first tested the effect of binding time between AuNUs and R6G, one of the important factors for effective SERS. The complete mechanism of SERS has not yet been revealed. However, there are two main mechanisms that have been widely accepted.28One is the electromagnetic (EM) effect, and the other is the chemical effect (CE). The EM effect is caused by an amplified EM field generated with superposition of the surface plasmon resonance of the gold nanoparticles. The CE is associated with the electronic coupling of molecules adsorbed on the surface and the charge transfer between the adsorbed analyte and the metal surface. Therefore, the binding time is related to the CE, having direct influence on the enhancement of SERS. We investigated the effective adsorption of R6G molecules at different binding times from 0 h (no binding time) to 8 h at room
temperature shown in figure 5A.29 For studying only the effect of binding time, we held the other factors constant for each sample. With no binding time, the SERS spectrum was rough and had barely present peaks of R6G, which was similar to the normal Raman data of the probe molecules in 1mM concentration. This means that spontaneous binding did not occur in a few minutes and took at least 2 h. If there were any successful adsorptions, there would be distinct Raman peaks indicative of the enhancement of SERS. The broad bands between 1200 cm−1and 1600 cm−1were caused by the capillary tubes and were not signals of the analytes. As the binding time increased, the enhancements increased, growing clear peaks from the R6G despite the interference of the capillary tubes. When the binding time was 6–8 h, the result was most prominent with the highest resolution, indicating the effective adsorption of molecules on the surface. Figure 5B shows Raman intensities at 1370 cm−1 at each condition in figure 5A, fitted with a linear function. As seen in figure 5B, the Raman intensity increased steadily with increasing binding time, and the point at 8 h was slightly off the linear fit line. However, after 8 h, the resolution was undesirable due to the weak interaction between the metals and the probe molecules.
R6G with its positive nitrogen site can interact with AuNUs through Van der Waals interactions, not by chemical bonding.30-31 In the CE, strong interactions lead to a high active SERS. However, the weak interaction between gold and R6G decreased the stability of the attached R6G, hence, the lifetime of this chemical form was shortened.32In other words, after a certain period, the R6G molecules could detach from the surface, which caused the decrease in resolution.
Figure 5. SERS spectra with increasing binding time from 0 h to 8 h (A), Raman intensities at 1370 cm−1 obtained from SERS spectra shown in (A), the experimental data were fitted with a linear function (B)
Next, we investigated the ratio between AuNUs and R6G by controlling metal concentrations while fixing the concentration of the probe molecules. Likewise, other conditions were kept constant during this experiment. We obtained the SERS spectra with 0.025 mM, 0.05 mM, and 0.075 mM of AuNUs while the concentration of R6G solution was fixed to 1 mM. With increasing concentrations of AuNUs, the SERS signals were enhanced noticeably with improved resolution shown in figure 6A.
Furthermore, the peaks at 1103 cm−1, 1140 cm−1, and 1197 cm−1, not observed in 0.025 mM, appeared at higher concentrations. For clearer comparison, we fitted the Raman intensities at 1370 cm−1 with a linear function as shown in figure 6B, and the experimental data were well fitted by the linear function. We then tried to find the limit for linear response by increasing the concentration continuously in figure 7. From the results, we observed that the Raman signal steadily increased with increasing concentration, and the greatest increase was observed at 0.3 mM (the ratio of metal to molecules was about 8). However, beyond that 0.3 mM concentration, the line collapsed, and the peaks were difficult to distinguish even though the overall spectra increased. This result can be ascribed to the excessive concentrations of metals inducing aggregation, causing the number of molecules participating in SERS to be reduced due to the narrower area of the surface to which the probe molecules could be attached.33As a result, the resolution was low because the system failed to have the minimum numbers of molecules required for enhancement. This is a reasonable explanation because we removed the surfactant to prevent the formation of aggregates between gold nanoparticles, which have a high probability of agglomeration.
Figure 6. SERS spectra at AuNUs concentrations from 0.025 mM to 0.075 mM.
Detailed data with more concentrations is provided in figure 7 (A), Raman intensities at 1370 cm−1obtained from SERS spectra shown in (A). the experimental data were fitted with a linear function (B)
Figure 7.SERS spectra at added AuNUs concentrations from 0.025 mM to 0.4 mM
During the SERS experiments, we noted that the magnitude of Raman enhancement differed slightly with the centrifugation conditions. The centrifugation step was carried out to remove the surfactants that surrounded the AuNUs, thereby minimizing the interference of the effective adsorption of R6G molecules on the particle surface. Because spontaneous desorption of surfactant from the surface is rare, we had to remove it intentionally, and centrifugation was the easiest and simplest method. We used four sets of centrifugation conditions: 2400 rcf for 20 min, 40 min and 4800 rcf for 20 min, 40 min. As shown in figure 8, centrifugation at 2400 rcf for 40 min showed greater enhancement for 100 nm AuNUs than did another centrifugation condition. This result means that the condition of 2400 rcf for 40 min can remove the surfactants from the surface more effectively so that the R6G
molecules can be more conveniently attached to the empty surface, resulting in the higher enhancement of SERS. Therefore, the centrifugation step has to be considered for effective SERS experiments and large Raman enhancement. In addition to the aforementioned centrifugation effect, we further investigated the stirring effect on SERS. Most sample preparation for SERS contained a stirring step for fast, complete adsorption in the system. With stirring, the adsorption time could be shortened, and the molecules could be attached more strongly to the metal surface, with additional energy generated from stirring of the samples. In the present study, we set the total binding time to 6–8 h and evaluated five different stirring times, from 0 h (meaning no stirring step) to 4 h. As shown in figure 9, we noticed that the longer stirring time degraded the intensity as well as the resolution. The spectrum of the sample stirred for 4 h rarely had peaks and showed the lowest intensities compared with other conditions. The red curve in figure 9, from the sample with no stirring, showed the distinct Raman peaks of R6G. The result can be explained by weak Van der Waals interactions between the R6G molecules and the AuNUs, not by chemical bonding.
This interaction can be influenced by a physical force such as stirring, and continuous stirring can easily interrupt the binding or remove any instances of weak binding. Therefore, we concluded that it is favored not to have a stirring step in sample preparations for SERS experiments using AuNUs and R6G molecules.
Figure 8.The effect of centrifugation on SERS enhancement
Figure 9.The stirring effect on SERS spectra during the binding of the molecules on the metal surface
III-2. SERS mechanism study focused on chemical effect III-2-1. Characterization of gold nanorods
AuNRs have an anisotropic shape and unique optical properties, unlike nanosphere. Figure 10A shows the identical shape and size of nanoparticles characterized by the SEM images and the size distribution data and additional SEM image are provided in figure 11 with the mean value, 86.94 (± 5.28) nm x 25.37 (±
2.48) nm and an aspect ratio as 3.5. The UV-Vis extinction spectrum of AuNRs in figure 10B indicates two localized surface plasmon resonance (LSPR) peaks derived from the shape effect. The left peak at 520 nm is a transverse mode generated with the short axis and the right one in NIR region, 776 nm is a longitudinal mode from the long axis of nanoparticles known as more sensitive.34-35 As abovementioned and designed by self-assembly monolayers (SAMs) probes from scheme 2, the dashed line near the LPSR longitudinal peak in figure 10B indicates that the laser power wavelength was set close to 770 nm. The latter enabled large hot electron populations to be transferred from AuNRs to probes molecules and favor a high probability of direct hot electrons transfer processes. Therefore, a resonance effect between longitudinal LSPR peak of AuNRs and a laser wavelength of 785 nm is expected in this study.
Figure 10.SEM image of AuNRs and a model with the average size 25 nm X 87 nm (A). UV-Vis spectrum of AuNRs showing two distinct LSPR peaks and the dotted- line indicated a resonance condition with a 785 nm laser (B)
Figure 11. Size distributions of AuNRs with gaussian fitting, measured average length 87 nm and width 25 nm. Total 90 particles are counted from SEM images
In fact, the case used off-resonance condition has been dominated for excluding the EM effect at most, concentrated with pure CE. However, we are willing to use the resonance condition because the EM effect has been identified as a main contribution toward enhancement.15When the EM effect got weakened by non- resonance condition, the magnitude of enhancement might damage with weak results
and low reproducibility. Even the spontaneous self-assembly monolayer system, not exquisitely-designed substrates for extreme enhancements, like our condition could go through troubles in obtaining enough data for effective comparison. For optimizing this experiment, we control the all factors influenced in EM effect with same condition and has gotten a same EM effect expecting high enhancement but focusing CE well.